HyperDynamics: Meta-Learning Object and Agent Dynamics with Hypernetworks

Authors: Zhou Xian, Shamit Lal, Hsiao-Yu Tung, Emmanouil Antonios Platanios, Katerina Fragkiadaki

ICLR 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We test Hyper Dynamics on a set of object pushing and locomotion tasks. It outperforms existing dynamics models in the literature that adapt to environment variations by learning dynamics over high dimensional visual observations, capturing the interactions of the agent in recurrent state representations, or using gradient-based meta-optimization. We evaluate Hyper Dynamics on singleand multi-step state predictions, as well as downstream model-based control tasks. Specifically, we apply it in a series of object pushing and locomotion tasks. Our experiments show that Hyper Dynamics is able to generate performant dynamics models that match the performance of separately and directly trained experts, while also enabling effective generalization to systems with novel properties in a few-shot manner.
Researcher Affiliation Academia Zhou Xian, Shamit Lal, Hsiao-Yu Fish Tung, Emmanouil Antonios Platanios & Katerina Fragkiadaki School of Computer Science, Carnegie Mellon University, Pittsburgh, PA 15213, USA
Pseudocode No The paper does not contain explicit pseudocode or algorithm blocks.
Open Source Code No The paper does not provide an explicit statement about releasing code or a link to a code repository for the methodology described.
Open Datasets Yes Our dataset consists of only 31 different object meshes with distinct shapes, including 11 objects from the MIT Push dataset (Yu et al., 2016) and 20 objects selected from four categories (camera, mug, bowl and bed) in the Shape Net Dataset (Chang et al., 2015).
Dataset Splits No The paper states: 'We split our dataset so that 24 objects are used for training and 7 are used for testing.' and 'All models are trained for 150 iterations till convergence. In addition, early stopping is applied if an rolling average of total return decreases.' While early stopping implies the use of a validation set, the paper does not explicitly describe the split of a validation set (e.g., percentages or counts).
Hardware Specification No The paper mentions 'Our setup uses a Kuka robotic arm equipped with a single rodshaped end-effector' as part of the experimental environment. However, it does not specify the computing hardware (e.g., CPU, GPU models, memory) used to run the experiments.
Software Dependencies No The paper mentions 'Py Bullet simulator (Coumans & Bai, 2016 2019)' and 'Mu Jo Co simulator (Todorov et al., 2012)' and uses various neural network architectures (e.g., GRUs, Mask R-CNN). However, it does not list specific software dependencies with version numbers (e.g., 'PyTorch 1.9').
Experiment Setup Yes For object pushing: 'We use batch size of 8 and a learning rate of 1e 3. We collected 50, 000 pushing trajectories for training, and 1, 000 trajectories for testing. All models are trained till convergence for 500K steps.' For locomotion: 'Here we use batch size of 128 and a learning rate of 1e 3. All models are trained for 150 iterations till convergence. In addition, early stopping is applied if an rolling average of total return decreases.'